Edinburgh Research Explorer
Scottish snow cover dependence on the North Atlantic
Oscillation index
Citation for published version: Spencer, M & Essery, R 2016, 'Scottish snow cover dependence on the North Atlantic Oscillation index', Hydrology research. https://doi.org/10.2166/nh.2016.085
Digital Object Identifier (DOI): 10.2166/nh.2016.085
Link: Link to publication record in Edinburgh Research Explorer
Document Version: Peer reviewed version
Published In: Hydrology research
General rights Copyright for the publications made accessible via the Edinburgh Research Explorer is retained by the author(s) and / or other copyright owners and it is a condition of accessing these publications that users recognise and abide by the legal requirements associated with these rights.
Take down policy The University of Edinburgh has made every reasonable effort to ensure that Edinburgh Research Explorer content complies with UK legislation. If you believe that the public display of this file breaches copyright please contact [email protected] providing details, and we will remove access to the work immediately and investigate your claim.
Download date: 25. Sep. 2021 1 Scottish snow cover dependence on the North
2 Atlantic Oscillation index
* * 3 Michael Spencer and Richard Essery
* 4 School of GeoSciences, University of Edinburgh, Grant Institute,
5 James Hutton Road, King’s Buildings, Edinburgh, EH9 3FE.
6 [email protected], 07851083738 (corresponding author)
8 Abstract
9 Forecasting seasonal snow cover is useful for planning resources and
10 mitigating natural hazards. We present a link between the North Atlantic
11 Oscillation (NAO) index and days of snow cover in Scotland between win-
12 ters beginning from 1875 to 2013. Using broad (5 km resolution), national
13 scale datasets like UK Climate Projections 2009 (UKCP09) to extract na-
14 tionwide patterns, we support these findings using hillslope scale data from
15 the Snow Survey of Great Britain (SSGB). Currently collected snow cover
16 data, are considered using remotely sensed satellite observations, from Mod-
17 erate Resolution Imaging Spectroradiometer (MODIS); but the results are
18 inconclusive due to cloud. The strongest correlations between the NAO in-
19 dex and snow cover are found in eastern and southern Scotland, these results
1 20 are supported by both SSGB and UKCP09 data. Correlations between NAO
21 index and snow cover are negative with the strongest relationships found for
22 elevations below 750 m. Four SSGB sites (two in eastern Scotland, two in
23 southern Scotland) were modelled linearly with resulting slopes between -6
24 and -16 days of snow cover per NAO index integer value. This is the first
25 time the relationship between NAO index and snow cover duration has been
26 quantified and mapped in Scotland.
27 Keywords: Climate, North Atlantic Oscillation, snow, Scotland
2 28 Introduction
29 Snow is important in Scotland for water resources, e.g.: the largest instrument-
30 measured flow in Scotland’s largest catchment, the River Tay, was partly caused
31 by snowmelt (Black and Anderson, 1994). Dunn et al. (2001) show that snow can
32 contribute to river baseflow until July, as melted snow takes a generally slower
33 sub-surface pathway to a water course. Also, Gibbins et al. (2001) discuss the
34 importance of snowmelt for freshwater invertebrate habitat in the Cairngorms.
35 Knowledge of snow extent and duration can help understand habitat change (Trivedi
36 et al., 2007), and global snow cover data are collated by the Intergovernmental
37 Panel on Climate Change (Vaughan et al., 2013).
38 The North Atlantic Oscillation index (NAO) is the normalised pressure dif-
39 ference between the Icelandic low and the Azores high (Walker and Bliss, 1932).
40 Positive NAO phases are typified by strong westerly winds carrying moist warm
41 air from the Atlantic, with negative NAO phases bringing colder air masses from
42 the east (Hurrell, 1995; Simpson and Jones, 2014). Logically then, the NAO in-
43 dex could indicate the duration of snow cover as colder weather means a greater
44 chance of snow and its persistence, but this signal may be confused by positive
45 NAO phases bringing increased precipitation.
46 NAO index relates to hydrological processes: Hannaford et al. (2005) show
47 river flow and NAO index have strong positive correlations (e.g. River Nith: 0.63)
48 in the north and west of the UK, but eastern catchments had a weaker correlation
49 (e.g. River Tweed: 0.38). Harrison et al. (2001) has suggested an association
50 between snow cover and NAO phase is likely. While Trivedi et al. (2007) found
51 snow cover, in the Ben Lawers region north of Loch Tay, at 300 m and below to be
52 significantly (p < 0.05) negatively correlated with NAO index, between -0.55 and
3 53 -0.38, with lower elevations having a stronger relationship. Trivedi et al. (2007)
54 also found no correlation between NAO index and falling snow, perhaps because
55 it is often cold enough for snow to fall during a Scottish winter, irrespective of
56 NAO phase, but during positive NAO phases the warmer air causes snow to melt
57 and only with the colder temperatures associated with negative NAO indices does
58 snow lie for longer. There has been more research on snow cover links to the NAO
59 index in continental Europe, where snow cover has a greater impact (e.g. Scherrer
60 et al. 2004; Kim et al. 2013; Bednorz 2004; Beniston 1997; Lopez-Moreno et al.
61 2011).
62 There has recently been an increase in winter variability of the NAO phase
63 (Hanna et al., 2014; Osborn, 2006), including a record low NAO index in 2009/10
64 (Osborn, 2010). The 2009/10 low occurred the same year as an exceptionally cold
65 and snowy winter in the UK (National Climate Information Centre, 2010; Prior
66 and Kendon, 2011). Goodkin et al. (2008) link variability in the NAO index to
67 northern hemisphere mean temperature and state any future predictions should
68 take this into account.
69 The UK Met Office are beginning to more successfully forecast seasonal NAO
70 indices (Scaife et al., 2014), which could be used to plan for heavy snow in ad-
71 vance of a winter season. For a forecast made on the 1st November Scaife et al.
72 (2014) give a correlation value of 0.62 (significant at 99 %) between forecast and
73 observed DJF NAO index for the years 1993 to 2012.
74 We hypothesise that snow cover in Scotland is negatively correlated to the
75 NAO index. We establish this by looking at nationwide snow cover datasets, be-
76 fore further investigating relationships at a hillslope scale, using case studies with
77 more detailed data available. Our paper is laid out as follows: methods and data,
4 78 results, discussion and conclusion. The methods and results sections are split by
79 dataset.
80 Data and methods
81 We used NAO index data from the Climate Research Unit, University of East
82 Anglia (Undated) and Osborn (Undated) as these comprise a long and definitive
83 record (Table 1). The longest data series of Scottish snow cover are from UK Met
84 Office stations which record snow presence at a given point at 09:00 hours UTC
85 each morning; the longest of these is Braemar which has recorded since 1927
86 (Harrison et al., 2001). 96 % of UK Met Office snow recording stations lie below
87 300 m elevation (Spencer et al., 2014) and so are unrepresentative of the 31 %
88 of Scottish landmass that is higher (Spencer et al., 2014). These UK Met Office
89 station data are used by proxy via the UK Climate Projections 2009 (UKCP09)
90 snow cover dataset (Met Office, Undated). Table 1 shows a non definitive list of
91 Scottish snow cover datasets, which are all used within this study.
92 Snow in Scotland is often ephemeral and so metrics like average snowline and
93 maximum snow cover extent are meaningless because each winter can see many
94 snow accumulation and melt cycles. We solved this by using a count of the days
95 of snow cover during a given time period. We define a winter period for snow
96 cover as November to April to help differentiate the snowiest winters, while being
97 short enough to not discount many SSGB records, as some are missing (Spencer
98 et al., 2014). A short winter period (e.g. DJF) would mean, particularly at higher
99 elevations, a count of days with snow lying would result in saturated values of
100 days of snow cover, i.e. there cannot be any more than 31 days with snow lying
101 in January, but 31 days of cover is often the case at higher elevations in Scotland.
5 102 Using a six month period will help identify the snowiest winters, where greater
103 snow depths take longer to melt. Analysis was undertaken using the R language
104 (R Core Team, 2015).
105 NAO
106 NAO index data have been averaged (mean) over DJFM, as described by Osborn
107 et al. (1999), to better represent the prevailing winter NAO index. Note this winter
108 period is different to the NDJFMA period used for snow cover. Fig. 1; shows the
109 predominant NAO index is positive, aligning with our understanding that the UK
110 is more likely to experience weather systems from the west.
111 Bonacina
112 The Bonacina snow index was originally compiled by Leo Bonacina (Bonacina,
113 1966) and is now maintained as a website (O’Hara and Bonacina, Undated). Each
114 winter is subjectively categorised into one of four groups: Little, Average, Snowy
115 and Very Snowy. This is based on how much snow fell and how much of the coun-
116 try it covered using anecdotal data from weather journals, UK Met Office stations
117 and websites. Other snow cover datasets used in this work state the number of
118 days of snow cover over a given time period. Bonacina data has been included
119 because it covers a much longer time period than the other snow cover datasets
120 (Table 1).
121 Mean DJFM NAO index values are grouped by Bonacina categories. The dif-
122 ferences between groups of the NAO index are compared visually using boxplots
123 (Fig. 2) and statistically using an ANOVA and Tukey honest significant differ-
124 ences (HSD) (Yandell, 1997) tests, the latter to account for family-wise analysis
6 125 (Table 2).
126 UK Climate Projections 2009 (UKCP09)
127 The UKCP09 snow dataset comprises a 5 km resolution raster image for each
128 month, where each grid value represents the number of days snow cover for that
129 cell. November to April data are available from 1971/72 until 2005/06. These
130 were interpolated from UK Met Office station data by Perry and Hollis (2005).
131 These data have been shown (Spencer et al., 2014) to poorly represent reality at
132 higher elevations. The dataset is used here to identify regions for more detailed
133 exploration. UKCP09 snow data were downloaded from Met Office (Undated).
134 The November to April sum of days of snow cover are compared using a Pearson
135 correlation to the mean DJFM NAO index. The resulting Pearson correlation is
136 plotted (Fig. 3) to show spatial patterns.
137 Snow Survey of Great Britain (SSGB)
138 The SSGB reported at 145 stations in Scotland at differing times between 1945
139 and 2007, but some records are missing (Spencer et al., 2014). Stations were
140 selected for inclusion in this study based on whether they recorded for all months
141 between November to April. The number of SSGB stations meeting this criterion
142 each year is shown in Fig. 4. The gaps in the number of reporting stations are
143 because data are missing from part of these years. This is directly related to only
144 including stations that recorded all months between November and April each
145 winter.
146 SSGB observers recorded the elevation of snowline on visible hillslopes sur-
147 rounding each station. We constructed snow accumulation curves, where the num-
7 148 ber of days snow cover over a range of elevations are shown. These accumulation
149 curves are split by NAO index and shown in Fig. 5. The primary purpose of these
150 curves is to assess the break point between higher and lower elevation snow cover.
151 Three groups of individual stations are also considered, again meeting the
152 criterion of six months of record for a winter; group one: stations with the longest
153 record, group two: stations in the east of Scotland, group three: a single station on
154 Orkney. Details of these stations are shown in Table 3 and the location in Fig. 6.
155 The second and third groups have much shorter records than the longest-running
156 stations, they have been included to help test whether eastern sites are more likely
157 to have snow cover influenced by the NAO index and whether the UKCP09 snow
158 snow data are a good approximation of snow cover. The groups of stations in
159 Table 3 are compared to the NAO index using a high and low elevation split (at
160 750 m) and a Loess (locally weighted scatterplot smoothing) (Cleveland, 1979;
161 Cleveland and Devlin, 1988) with 95 % confidence limits (Fig. 7 and 8).
162 Stations from Table 3, judged by eye to have a Loess close to a straight line,
163 are plotted in Fig. 9 with linear models, showing the Pearson correlation value and
164 line parameters (slope and intercept). This allows us to relate a given NAO index
165 to an expected number of days snow cover duration for a high or low elevation.
166 Moderate Resolution Imaging Spectroradiometer (MODIS)
167 There are two main methods for remote sensing of snow; microwave and visible.
168 Using microwave to detect snow cover is very challenging in mountainous terrain
169 (Snehmani et al., 2015) or when snow is wet (Rees and Steel, 2001). Snehmani
170 et al. (2015) review methods that improve microwave assessment of snow cover,
171 but these are data and computing intensive and trialling them in Scotland where
8 172 it is very cloudy, wet and mountainous is beyond the scope of this study. Some
173 snow cover datasets amalgamate different data sources, including Robinson et al.
174 (Undated) and Foster et al. (2011), which have grid resolutions of 190.5 km and
175 25 km respectively; these are coarse grids which would miss spatial detail. Foster
176 et al. (2011) found that Earth Observation System Moderate Resolution Imaging
177 Spectroradiometer (MODIS) outperformed microwave snow detection in cloud
178 free areas. MODIS is freely available on a 500 m grid at a twice daily resolution,
179 and there are some reanalysis products, e.g. Notarnicola et al. (2013), which re-
180 calculate snow cover at a 250 m grid, but are only available for the Alps. MODIS
181 data are used in this study because of the temporal overlap with SSGB data and
182 fine resolution of the dataset. The MODIS dataset chosen was the tile set which
183 records as binary whether snow covered each cell, rather than the fractional or
184 albedo datasets. Coverage of Scotland is split across two tiles; these were down-
185 loaded from the National Snow and Ice Data Centre (Hall et al., 2006) for both the
186 Aqua (2002-07-04 onwards) and Terra (2000-02-24 onwards) satellites. Each pair
187 of tiles were merged together and reprojected to the British National Grid using
188 GDAL (GDAL Development Team, 2015). Using GRASS GIS software (GRASS
189 Development Team, Undated), a combination of both satellites was created to re-
190 duce the incidence of cloud pixels by approximately 15 %. This method was only
191 possible from 2002-07-04 onwards, when the Aqua satellite became operational.
192 Prior to this the Terra satellite alone was used, creating a dataset containing full
193 winters from 2000/01 until 2013/14. These November to April period data were
194 summed and correlated against the DJFM NAO mean index, presented in Fig. 10a.
195 Fig. 10b shows the equivalent analysis, but repeated for cloud cover observed by
196 MODIS.
9 197 Data comparison
198 To relate SSGB station and national results, Pearson correlations from SSGB,
199 MODIS and UKCP09 are compared. Values from MODIS and UKCP09 rasters
200 were extracted at SSGB station locations and are shown together in Table 4.
201 Results
202 Bonacina
203 Fig. 2 shows boxplots of the difference between DJFM NAO index as grouped
204 by the Bonacina classification. A general trend can be seen where less snowy
205 winters have a more positive NAO index. This is demonstrated statistically using
206 ANOVA (F value = 25.07) and a Tukey HSD analysis (Table 2) where each pair
207 is shown with a best estimate of difference and significance value. All pairs are
208 different at greater than 5 % significance, except Very Snowy - Snowy. This could
209 be a product of the Very Snowy small sample size, where the Tukey HSD test
210 performs less well.
211 UKCP09 snow
212 Fig. 3 shows some strongly negatively correlated areas of Scotland. The strongest
213 correlations are in the south west and along the east coast. Areas of poor correla-
214 tion are predominantly in central and northern mainland Scotland and Orkney.
10 215 SSGB
216 Fig. 5, showing SSGB snow accumulation curves, displays a marked difference
217 in duration of snow cover at all elevations between winters with the highest and
218 lowest NAO indices, with positive NAO phases having less snow cover than neg-
219 ative NAO phases. Below 750 m the difference in days of snow cover as elevation
220 increases are broadly linear, while above 750 m the relationship is unclear, with
221 lines crossing. This 750 m change-point is used to distinguish between high and
222 low snow cover for the SSGB station analysis.
223 Individual SSGB stations with the longest record of complete winters and
224 some other stations are considered (Table 3). Other stations, in the east and
225 Orkney, were used to investigate the more extreme correlations between the NAO
226 index and UKCP09 snow data (Fig. 3), accepting that they do not have the longest
227 records. These results corroborate what is shown in the UKCP09 snow results
228 (Fig. 3), that south western sites like Forrest Lodge (Fig. 7) show a negative
229 correlation with the NAO index. This is repeated in Fig. 8 where eastern sites,
230 Crathes and Whitehillocks, show a strong relationship to the NAO index. Also in
231 line with the UKCP09 results, Stenness, chosen because of a poor UKCP09 snow
232 correlation to the NAO index, shows a weak relationship to NAO index (Fig. 8).
233 SSGB stations Crathes, Eskdalemuir, Forrest Lodge and Whitehillocks have
234 been plotted with a linear regression line, Fig. 9. Line slopes vary from -7 to
235 -14 days for higher elevations and from -6 to -16 days for lower elevations. As
236 can be seen in Fig. 5 to 8 the NAO index has a larger impact at lower elevations,
237 but Pearson correlation values are variable; this could be a function of stations
238 not observing the same time periods and hence some sampling produces better
239 correlations than others. None of the SSGB stations observed the record NAO
11 240 index low of winter 2009/10.
241 MODIS
242 Fig. 10 was interpolated (bilinear) to a 5 km resolution, to better show correla-
243 tions. Fig. 10a shows a generally weak correlation between MODIS snow cover
244 and the NAO index. The strongest correlations are in north west Scotland, with
245 the weakest in central eastern Scotland. Orkney shows a strong correlation, in
246 contrast to the UKCP09 and SSGB results. A small proportion of the plot, east of
247 Edinburgh, has a very weak but positive correlation, in disagreement with Fig. 2
248 to 9.
249 Differences from UKCP09 and SSGB results are most likely because of the
250 frequency of cloud, as it is difficult for visible remote sensing to see through cloud.
251 The problem is illustrated in Fig. 10b, which shows cloud cover; as interpreted by
252 MODIS, correlated with the NAO index. The area of positive correlation exceeds
253 the area of negative correlation. An east-west split in correlation is clearly shown,
254 with the east coast negatively correlated to the NAO index and the west coast
255 positively correlated to the NAO index. This will have an impact on seeing spatial
256 snow cover trends; if we expect the east to get more days of snow cover when there
257 is a negative NAO index, a corresponding increase in cloud cover will obscure
258 snow observations.
259 Data comparison
260 A comparison of correlations from different datasets can be seen in Table 4. These
261 results are summarised by Pearson correlations between datasets, UKCP09: 0.87
262 and MODIS: -0.07, demonstrating that the SSGB and UKCP09 results corroborate
12 263 each other, but that MODIS results do not correlate with SSGB results.
264 Discussion
265 There is a strong correlation between UKCP09 and SSGB results, with high-
266 lighted areas like south west Scotland and east Scotland showing strong negative
267 correlations between snow cover and the NAO index and Orkney with no corre-
268 lation. This indicates that UKCP09 is an appropriate method for analysing the
269 spatial relationship between snow cover and NAO phase at a national scale. The
270 SSGB data have shown the stronger correlation between the NAO index and snow
271 cover at lower elevations. We believe this is because lower elevations have more
272 transient snow as they are generally warmer than higher elevations and so snow
273 will be less likely to fall and lying snow will more readily melt. This makes snow
274 in these areas susceptible to even small changes in temperature. Perhaps most im-
275 portantly: the persistence of snow at lower elevations is less, because increases in
276 temperature from westerly air flows have a greater impact on areas that are closer
277 to melt. This low elevation correlation is supported, by proxy, by the Bonacina
278 index correlation to the NAO index, Fig. 2, as the majority of Great Britain is low
279 lying, so the Bonacina index is more likely to reflect the more common (lower)
280 elevation zone than more remote mountain areas. Our correlations of NAO index
281 and snow cover are weaker for higher elevations, which are often cold enough
282 for deeper snow to accumulate and take longer to melt for a wider range of typ-
283 ical winter temperatures. The most recent example of this was winter 2013/14,
284 which was comparatively mild and very wet, but vast quantities of snow fell at
285 higher elevations in Scotland (Kendon and McCarthy, 2015). Kendon and Mc- ◦ 286 Carthy (2015) discuss a lapse rate of approximately 6 C/km between Aviemore
13 287 and Cairngorm summit, which was linked to the persistent Atlantic weather type
288 and absence of temperature inversions. This lapse rate is higher than the long- ◦ 289 term (1983 to 2008) average of 5.2 C/km for Aviemore and Cairngorm chair lift
290 calculated by Burt and Holden (2010), helping to explain the depth and duration
291 of snow cover accumulated that winter.
292 Inland areas generally have a poorer correlation with the NAO index. As much
293 of this area is high in elevation this can partly be attributed to it being cold enough
294 for snow to accumulate and persist, irrespective of the NAO index. These con-
295 tinental areas may also be dominated more by local weather systems and micro-
296 climates, enabling snow to persist for longer.
297 Those stations that showed a more easily defined relationship with a Loess
298 have had linear models fitted (Fig. 9), with Pearson correlation values, from -0.29
299 to -0.5. This range of results could be explained by micro-climates having a bigger
300 impact on snow cover than long-term weather patterns. This would be especially
301 true on the east side of the Cairngorms, where wind (predominantly westerly)
302 driven snow often accumulates on eastern slopes and can take a long time to melt.
303 These spatial local discrepancies can also be temporal, given that the SSGB sites
304 did not all observe the same winters, some may have been more closely correlated
305 with the NAO index than others. The obvious solution is to consider the results
306 from Fig. 5, which average over a greater number of SSGB stations, helping to
307 reduce uncertainty.
308 Conclusion
309 Spatial variability of snow cover is a big challenge, it is difficult to observe and
310 quantify. This is typified by the contrasting results of UKCP09 snow and MODIS
14 311 data correlations. We have overcome this by using disparate snow cover datasets,
312 encompassing anecdotal type data (Bonacina index), interpolated ground observed
313 data (UKCP09), the SSGB and satellite observations (MODIS). With the excep-
314 tion of the MODIS analysis, these have all shown the same results: that Scottish
315 snow cover is generally negatively correlated with the NAO index, with stronger
316 correlations at lower elevations and in southern and eastern Scotland. Results
317 from individual SSGB stations and UKCP09 grids correlate well demonstrating
318 the value of UKCP09 data for national scale assessment of spatial trends. At sam-
319 ple locations, increases in snow lying between November and April are 6 to 16
320 days for each unit reduction in NAO index. These estimates could be used in
321 conjunction with seasonal NAO forecasts in preparation for upcoming winters by
322 groups like highways and local authority planners and snow sports industries.
323 As new snow datasets become available, particularly from satellite and reanal-
324 ysis products, it will be worthwhile revisiting and updating this research to help
325 constrain uncertainty. This will be particularly pertinent if predictions of a more
326 volatile NAO index come to pass, as we will be able to better link snow cover
327 to climate variability, helping our understanding of snow cover in a changing cli-
328 mate.
15 329 References
330 Bednorz, E. Snow cover in Eastern Europe in relation to temperature, precipitation
331 and circulation. Internaltional Journal of Climatology, 24(5):591–601, 2004.
332 Beniston, M. Variations of snow depth and duration in the swiss alps over the
333 last 50 years: links to changes in large-scale climatic forcings.(special issue:
334 Climatic change at high elevation sites). Climatic Change, 36:281–300, 1997.
335 Black, A. and Anderson, J. The great Tay flood of January 1993. Hydrological
336 Data UK: 1993 Yearbook, pages 25–34, 1994.
337 Bonacina, L. C. W. Chief events of snowfall in the British Isles during the decade,
338 1956-65. Weather, 21(2):42, 1966.
339 Burt, T. and Holden, J. Changing temperature and rainfall gradients in the british
340 uplands. Climate Research, 45(1):57–70, 2010.
341 Cleveland, W. S. Robust locally weighted regression and smoothing scatter plots.
342 Journal of Econometrics, 74:87–114, 1979.
343 Cleveland, W. S. and Devlin, S. J. Locally weighted regression: An approach to
344 regression analysis by local fitting. Journal of the American Statistical Associ-
345 ation, 83(403):596–610, 1988.
346 Climate Research Unit, University of East Anglia. North Atlantic Oscillation
347 data (1821-2000). http://www.cru.uea.ac.uk/cru/data/nao/, Undated. Accessed:
348 2015-03-13.
349 Dunn, S., Langan, S., and Colohan, R. The impact of variable snow pack accu-
16 350 mulation on a major Scottish water resource. Science of the Total Environment,
351 265:181–194, 2001.
352 Foster, J. L., Hall, D. K., Eylander, J. B., Riggs, G. A., Nghiem, S. V., Tedesco,
353 M., Kim, E., Montesano, P. M., Kelly, R. E. J., Casey, K. A., and Choudhury, B.
354 Blended global snow product using visible, passive microwave and scatterom-
355 eter satellite data. International journal of remote sensing, 32(5):1371–1395,
356 2011.
357 GDAL Development Team. GDAL - Geospatial Data Abstraction Library, Version
358 1.10.1. Open Source Geospatial Foundation, 2015. URL http://www.gdal.
359 org.
360 Gibbins, C., Dilks, C., Malcolm, R., Soulsby, C., and Juggins, S. Invertebrate
361 communities and hydrological variation in Cairngorm mountain streams. Hy-
362 drobiologia, 462(1-3):205–219, 2001.
363 Goodkin, N. F., Hughen, K. A., Doney, S. C., and Curry, W. B. Increased mul-
364 tidecadal variability of the North Atlantic Oscillation since 1781. Nature Geo-
365 science, 1:844–848, 2008.
366 GRASS Development Team. Geographic Resources Analysis Support System
367 (GRASS) software. http://grass.osgeo.org, Undated.
368 Hall, D. K., Riggs, G. A., and Salomonson, V. V. Updated daily. MODIS Terra
369 and Aqua snow cover daily L3 global 500m grid V005, 2000-2014. Digital
370 Media., 2006. Boulder, Colorado USA: National Snow and Ice Data Center.
371 Hanna, E., Cropper, T. E., Jones, P. D., Scaife, A. A., and Allan, R. Recent sea-
372 sonal asymmetric changes in the NAO (a marked summer decline and increased
17 373 winter variability) and associated changes in the AO and Greenland Blocking
374 Index. International Journal of Climatology, 35(9):2540–2554, 2014.
375 Hannaford, J., Laize, C., and Marsh, T. An assessment of runoff trends in undis-
376 turbed catchments in the Celtic regions of north west Europe. Number ISBN:
377 978-1-901502-88-6. IAHS Publication 310, 2005.
378 Harrison, S., Winterbottom, S., and Johnson, R. A preliminary assessment of the
379 socio-economic and environmental impacts of recent changes in winter snow
380 cover in Scotland. Scottish Geographical Journal, 117(4):297–312, 2001.
381 Hurrell, J. W. Decadal trends in the North Atlantic Oscillation: regional tempera-
382 tures and precipitation. Science, 269(5224):676–679, 1995.
383 Kendon, M. and McCarthy, M. The UK’s wet and stormy winter of 2013/2014.
384 Weather, 70(2):40–47, 2015.
385 Kim, Y., Kim, K.-Y., and Kim, B.-M. Physical mechanisms of European winter
386 snow cover variability and its relationship to the NAO. Climate Dynamics, 40
387 (7-8):1657–1669, 2013.
388 Lopez-Moreno, J., Vicente-Serrano, S., Moran-Tejeda, E., Lorenzo-Lacruz, J.,
389 Kenawy, A., and Beniston, M. Effects of the North Atlantic Oscillation (NAO)
390 on combined temperature and precipitation winter modes in the Mediterranean
391 mountains: Observed relationships and projections for the 21st century. Global
392 and Planetary Change, 77(1-2):62–76, 2011.
393 Met Office. UKCP09: Download data sets.
394 http://www.metoffice.gov.uk/climatechange/science/monitoring/ukcp09/download/index.html,
395 Undated. Accessed: 2015-03-13.
18 396 National Climate Information Centre. UK seasonal weather summary – winter
397 2009/2010. Weather, 65(4):99–99, 2010.
398 Notarnicola, C., Duguay, M., Moelg, N., Schellenberger, T., Tetzlaff, A., Mon-
399 sorno, R., Costa, A., Steurer, C., and Zebisch, M. Snow cover maps from
400 MODIS images at 250 m resolution, part 1: Algorithm description. Remote
401 Sensing, 5(1):110–126, 2013.
402 O’Hara, D. and Bonacina, L. C. W. British win-
403 ter snowfall events 1875-2014 (Bonacina/O’Hara).
404 http://www.neforum2.co.uk/ferryhillweather/bonacina.html, Undated. Ac-
405 cessed: 2015-03-13.
406 Osborn, T. Tim Osborn: North Atlantic Oscillation index data (1999 onwards).
407 http://www.cru.uea.ac.uk/ timo/datapages/naoi.htm, Undated. Accessed: 2015-
408 03-13.
409 Osborn, T. J., Briffa, K. R., Tett, S. F. B., Jones, P. D., and Trigo, R. M. Evalua-
410 tion of the North Atlantic Oscillation as simulated by a coupled climate model.
411 Climate Dynamics, 15(9):685–702, 1999.
412 Osborn, T. J. Recent variations in the winter North Atlantic Oscillation. Weather,
413 61(12):353–355, 2006.
414 Osborn, T. Winter 2009/2010 temperatures and a record-breaking North Atlantic
415 Oscillation index. Weather, 66(1):19–21, 2010.
416 Perry, M. and Hollis, D. The generation of monthly gridded datasets for a range
417 of climatic variables over the UK. International Journal of Climatology, 25(8):
418 1041–1054, 2005.
19 419 Prior, J. and Kendon, M. The UK winter of 2009/2010 compared with severe
420 winters of the last 100 years. Weather, 66(1):4–10, 2011.
421 R Core Team. R: A Language and Environment for Statistical Computing.R
422 Foundation for Statistical Computing, Vienna, Austria, 2015. URL http://
423 www.R-project.org.
424 Rees, W. and Steel, A. Radar backscatter coefficients and snow detectability for
425 upland terrain in Scotland. International Journal of Remote Sensing, 22(15):
426 3015–3026, 2001.
427 Robinson, D., Estilow, T., and NOAA CDR Program. NOAA Climate Date
428 Record (CDR) of Northern Hemisphere (NH) Snow Cover Extent (SCE), Ver-
429 sion 1., Undated. Accessed: 2015-03-22.
430 Scaife, A. A., Arribas, A., Blockley, E., Brookshaw, A., Clark, R. T., Dunstone,
431 N., Eade, R., Fereday, D., Folland, C. K., Gordon, M., Hermanson, L., Knight,
432 J. R., Lea, D. J., MacLachlan, C., Maidens, A., Martin, M., Peterson, A. K.,
433 Smith, D., Vellinga, M., Wallace, E., Waters, J., and Williams, A. Skillful
434 long-range prediction of European and North American winters. Geophysical
435 Research Letters, 41(7):2514–2519, 2014.
436 Scherrer, S., Appenzeller, C., and Laternser, M. Trends in Swiss Alpine snow
437 days: The role of local- and large-scale climate variability. Geophysical Re-
438 search Letters, 31(13), 2004.
439 Simpson, I. and Jones, P. Analysis of UK precipitation extremes derived from Met
440 Office gridded data. International Journal of Climatology, 34(7):2438–2449,
441 2014.
20 442 Snehmani, Singh, M. K., Gupta, R., Bhardwaj, A., and Joshi, P. K. Remote sens-
443 ing of mountain snow using active microwave sensors: a review. Geocarto
444 International, 30(1):1–27, 2015.
445 Spencer, M., Essery, R., Chambers, L., and Hogg, S. The historical snow survey
446 of Great Britain: Digitised data for Scotland. Scottish Geographical Journal,
447 130(4):252–265, December 2014.
448 Trivedi, M., Browne, M., Berry, P., Dawson, T., and Morecroft, M. Projecting
449 climate change impacts on mountain snow cover in central Scotland from his-
450 torical patterns. Arctic Antarctic and Alpine Research, 39(3):488–499, 2007.
451 Vaughan, D., Comiso, J., Allison, I., Carrasco, J., Kaser, G., Kwok, R., Mote,
452 P., Murray, T., Paul, F., Ren, J., Rignot, E., Solomina, O., K., S., and Zhang,
453 T. Observations: Cryosphere. In: Climate Change 2013: The Physical Sci-
454 ence Basis. Contribution of Working Group I to the Fifth Assessment Report of
455 the Intergovernmental Panel on Climate Change. Cambridge University Press,
456 Cambridge, United Kingdom and New York, NY, USA., 2013.
457 Walker, G. and Bliss, E. World weather v. Memoirs of the Royal Meteorological
458 Society, 4(36):53–84, 1932.
459 Yandell, B. S. Practical Data Analysis for Designed Experiments. Chapman and
460 Hall, 1997.
21 Table 1: Study data sources. Name Abbreviation Reference Type Time span Bonacina Bonacina O’Hara and Classification of 1875 onwards snowfall Bonacina snowiness of UK catalogue (Undated) winter UK Climate UKCP09 Perry and Hol- Interpolated grid 1971 - 2006 Projections lis (2005) of UK Met Office 2009 snow station data (days lying grid per month) MODIS satel- MODIS Hall et al. Daily classified 2000 onwards lite snow (2006) raster image cover, daily l3 500m grid v005 North Atlantic NAO index Osborn (Un- Single annual 1821 onwards Oscillation In- dated) value (DJFM dex mean) Snow Sur- SSGB (Spencer et al., Daily observa- 1945 - 2007 vey of Great 2014) tions of snowline Britain elevation
22 Table 2: Tukey HSD difference in medians of NAO indicies between pairs of Bonacina classes. Pair Difference P value Very Snowy-Snowy -0.823 0.093 Snowy-Average -0.670 0.008 Average-Little -0.697 0.002
23 Table 3: Longest, eastern and Orkney SSGB stations details. Station Easting Northing Description Complete winters Eskdalemuir 323500 602600 Longest 46 Couligarton 245400 700700 Longest 44 Forrest Lodge 255500 586600 Longest 44 Ardtalnaig 270200 739400 Longest 39 Fersit 235100 778200 Longest 39 Drummuir 337200 844100 eastern 24 Derry Lodge 303600 793200 eastern 21 Crathes 375800 796900 eastern 20 Whitehillocks 344860 779790 eastern 27 Stenness 329800 1011200 Orkney 21
24 Table 4: Pearson correlations of snow cover and NAO at SSGB stations with geo- graphically corresponding values extracted from MODIS and UKCP09 rasters. Station Elevation SSGB UKCP09 MODIS Ardtalnaig high -0.20 -0.41 -0.40 Ardtalnaig low -0.27 -0.41 -0.40 Couligarton high -0.18 -0.30 -0.34 Couligarton low -0.10 -0.30 -0.34 Crathes low -0.43 -0.52 -0.33 Crathes high -0.37 -0.52 -0.33 Derry Lodge low -0.23 -0.22 -0.53 Derry Lodge high -0.13 -0.22 -0.53 Drummuir high -0.52 -0.46 -0.53 Drummuir low -0.52 -0.46 -0.53 Eskdalemuir high -0.38 -0.49 -0.30 Eskdalemuir low -0.38 -0.49 -0.30 Fersit low -0.11 -0.27 -0.53 Fersit high -0.25 -0.27 -0.53 Forrest Lodge low -0.29 -0.51 -0.48 Forrest Lodge high -0.32 -0.51 -0.48 Stenness high 0.02 -0.05 -0.51 Stenness low 0.02 -0.05 -0.51 Whitehillocks high -0.41 -0.55 -0.54 Whitehillocks low -0.50 -0.55 -0.54
25 a) Mean NAO index (DJFM) index Mean NAO −4 −2 0 2 4
1820 1839 1858 1877 1896 1915 1934 1953 1972 1991 2010
Winter beginning
b) Frequency 0 5 10 15 20 25
−3 −2 −1 0 1 2 3
NAO index (mean DJFM)
Figure 1: Mean DJFM NAO index shown: a) through time, b) as a histogram. Positive index: red, negative index: blue.
26 Very Snowy n=8
Snowy n=38
Average n=39
Little n=54
−2 −1 0 1 2 3
Mean NAO index (DJFM)
Figure 2: Boxplots (median, upper and lower quartiles and range) showing winter NAO index grouped by Bonacina snowiness categories.
27 0.0
−0.1
−0.2
−0.3
Northing (m) −0.4 Pearson correlation Pearson
−0.5
−0.6 600000 700000 800000 900000 1000000 1100000 1200000
100000 200000 300000 400000 500000
Easting (m)
Figure 3: Map of Pearson correlation values between UKCP09 snow and the NAO index. Contains Met Office data c Crown copyright and database right 2015.
28 Stations 0 10 20 30 40 50 60
1946 1952 1958 1964 1970 1976 1982 1988 1996 2002
Winter beginning
Figure 4: Number of SSGB stations each year recording all six months between November and April.
29 NAO index = −2, n = 77 NAO index = −1, n = 429 NAO index = 0, n = 652 NAO index = 1, n = 382 NAO index = 2, n = 179 NAO index = 3, n = 33 Days with snow lying (Nov−Apr) with snow Days 0 50 100 150
0 200 400 600 800 1000 1200
Elevation (m)
Figure 5: Snow cover duration curves derived from SSGB data between 1946 and 2006 (Nov to Apr), grouped by (rounded) mean DJFM NAO index.
30 StennessStenness
1200
1000 DrummuirDrummuir
DerryDerry LodgeLodge 800 CrathesCrathes FersitFersit WhitehillocksWhitehillocks
Northing (m) 600 ArdtalnaigArdtalnaig
CouligartonCouligarton CouligartonCouligarton m AOD Elevation 400
200 EskdalemuirEskdalemuir Forrest Lodge 600000 700000 800000 900000 1000000 ForrestForrest LodgeLodge
100000 200000 300000 400000
Easting (m)
Figure 6: SSGB station locations. Contains Ordnance Survey data c Crown copy- right and database right 2015.
31 Ardtalnaig Couligarton Eskdalemuir Fersit Forrest Lodge 200 ● ● ● ● ● ●● ● ●● ●●● ● ● ●●●● ● ● ● ●● ● ●●●●●● ●● ● ●●● ● High elevation ● ● ● ● ● ●● ● ● ●● ● ● ● ● ●● ●● 150 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ●●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● 100 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ●●● ● ● ●● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●● ●●● ● ●●● ●● ● ● ● ● ●●● ●●● ● ● ● ●● ●● ●●● ●● ● ●● ● ●● ● ● ●● ● ● ● 50 ● ●●●● ●● ● ●●● ●●●● ● ● ●●● ● ● ● ●● ● ● ● ● ● ●● ● ● ● 0 ●
● ● ● 150 ● ● ● ● ● Low elevation ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● 100 ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●● ● ●● ● ● ● ●● ●●●● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ●● ●● ● ● ● ●●● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ●● ●● ● ● ● ● ● ● ● ● ● ●● ● ●●● ●● ●● 50 ●●●● ●● ● ●● ● ● ●●● ● ● ● ●● ● ● ● ● ● ●●●● ●● ● ●● ● ●●●●●● ● ● ● ● ●● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● Days snow cover per year (Nov − Apr) (Nov per year cover snow Days 0 ● −2 −1 0 1 2 3 −2 −1 0 1 2 3 −2 −1 0 1 2 3 −2 −1 0 1 2 3 −2 −1 0 1 2 3 North Atlantic Oscillation index (mean DJFM)
Figure 7: Long-record SSGB stations snow cover plotted against the NAO index, shown with a Loess and 95 % confidence bounds.
32 Crathes Derry Lodge Drummuir Stenness Whitehillocks
200
●● ● ●●● ● High elevation ●● ●● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● 150 ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ●● 100 ● ●● ●●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● 50 ● ● ●● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ●● ●● ● ● ● ● ● ●● ● ● ● ●● ●● ● ● ● ●● 0 ● ● ● ● 200
● ● Low elevation ● ●● 150 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ● 100 ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ●● ● ● ●●●● ● ● ● ● ● ● ●● 50 ● ● ● ●●● ●● ● ● ● ● ● ● ●● ● ● ● ● ●● ●● ● ● ● ● ● ● ● Days snow cover per year (Nov − Apr) (Nov per year cover snow Days ● ●● ● ● ● ● ●● ●● ● ● ● ●● ● 0 ● ● ● ● ● −2 −1 0 1 2 3 −2 −1 0 1 2 3 −2 −1 0 1 2 3 −2 −1 0 1 2 3 −2 −1 0 1 2 3 North Atlantic Oscillation index (mean DJFM)
Figure 8: Eastern and Orkney SSGB stations snow cover plotted against the NAO index, shown with a Loess and 95 % confidence bounds.
33 Crathes Eskdalemuir Forrest Lodge Whitehillocks
●●
● ● ● ● ● 150 ● Correlation = −0.38 Correlation = −0.32 ● ● ● Slope = −8 Slope = −7 ● ● High elevation ● ● ● ● ● ● ● ● ● ● ● Intercept = 55 ● Intercept = 61 ● ● ● ● ● ● ● ● ● ● ● ● ● 100 ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●● ● ● ● ●● ● ●● ● ● ●● ● ● ● ● ● 50 ● ● ●● ● ● ●● ● ● ●● ●● ● ●●●● ●● ● ● Correlation = −0.37 ● ●● ● ● ● ● Correlation = −0.41 ● ● ● ● ● ● ● ● ● Slope = −12 ● Slope = −14 ●● ● 0 Intercept = 96 ● Intercept = 119 150 Correlation = −0.38 Correlation = −0.29 ● Slope = −8 Slope = −6 ● ● ● Intercept = 55 Intercept = 53 ●
● ● ●● Low elevation ● ● ● ● ● ● ● ● ● ● ● 100 ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● Days snow cover per year (Nov − Apr) (Nov per year cover snow Days ● ● ● ●● 50 ● ● ● ● ● ●● ● ● ●●● ● ● ● ● ● ● ● ● ● ●● ● ●●● ●● ● ● ●● ● ● ● ● Correlation = −0.43 ● ● ● ● Correlation● = −0.5 ● ● ● ● ● ● ● ● ● ● ● ● ● Slope = −12 ● Slope = −16 ●● ● ● ● ● Intercept = 79 ● Intercept = 82 0 −2 −1 0 1 2 3 −2 −1 0 1 2 3 −2 −1 0 1 2 3 −2 −1 0 1 2 3 North Atlantic Oscillation index (mean DJFM)
Figure 9: Comparison between days snow cover at select SSGB stations in years that reported all months between November and April and the NAO index. Shown with a linear model with 95 % confidence bounds and a Loess smoother (dark grey) for comparison.
34 a) b)
1200000 0.6
1100000 0.4
1000000 0.2
900000 0.0
800000 −0.2 Northing (m)
−0.4 700000
−0.6 600000
−0.8 100000 200000 300000 400000 Pearson Correlation Easting (m)
Figure 10: a) Correlation between number of days MODIS recorded snow cover each winter (Nov to Apr) and the mean DJFM NAO index. b) Correlation between number of days MODIS recorded cloud cover each winter (Nov to Apr) and the mean DJFM NAO index.
35